Clinical trials, which test the safety and efficacy of drugs in a controlled population, cannot identify all safety issues associated with drugs because the size and characteristics of the target population, duration of use, the concomitant disease conditions and therapies differ markedly in actual usage conditions. On the outpatient side, medication related morbidity and mortality in the United States is estimated to result in 100,000 deaths and $177 billion in cost annually. On the inpatient side, it is estimated that roughly 30% of hospital stays have an adverse drug event. Current one-drug-at-a-time methods for surveillance are woefully inadequate because no one monitors the real life situation of patients getting over 3 concomitant drugs in the context of multiple co-morbidities. In preliminary work, we have built an annotation and analysis pipeline that uses the knowledge-graph formed by public biomedical ontologies for the purpose data-mining unstructured clinical notes. We have demonstrated that we can reproduce drug safety signals from the clinical notes on average 2.7 years ahead of the issue of a drug safety alert. Using this pipeline, we propose: 1) to identfy and prioritize multi-drug combinations that are worth testing; 2) to develop methods for discovering adverse event profiles of multi- drug combinations; and 3) to create an EHR derived catalogue of potential adverse events of multi-drug combinations. We will use hierarchies provided by existing public ontologies for drugs, diseases and side- effects to improve signal detection by aggregation, to reduce multiple hypothesis testing and to make a search for multi-drug side effects computationally tractable.